In image processing carried out by linking images captured by a plurality of cameras, it is important to control, with high accuracy, relationships between three dimensional (3D) positions and orientations (attitudes) of the cameras. Also, the relationships between 3D positions and orientations of the cameras change depending on impact, temperature, lapse of time, and so on. Therefore, control parameters of these relationships need to be updated regularly.
Correction of images captured by cameras is known as an approach to controlling relationships between 3D positions and orientations of cameras such that given relationships are present. As an approach to determining parameters used for image correction, i.e., a camera calibration method, for example, correction parameters are found such that desired images are produced as a result of capture of a specific pattern with the respective cameras. This method requires shooting of a specific pattern, making it difficult to update the parameters regularly.
For this reason, there is proposed a method of determining parameters such that similarity of a known object in each image is enhanced by using images captured by a plurality of cameras under normal use without using a specific pattern (refer, for example, to PTL 1). This method permits determination of parameters under normal use, thereby allowing for regular updating of the parameters. Also, there is conceived a method of determining parameters by shooting an arbitrary scene.